A Hybrid Wavelet Decomposition and Multi-Head Attention Bidirectional LSTM Framework for Short-Term Load Forecasting in Smart Grids
Keywords:
Short-term load forecasting, Smart grid, Wavelet decomposition, Bidirectional LSTM, Multi-head attention, Temporal Convolutional NetworkAbstract
Accurate short-term load forecasting (STLF) is fundamental to smart grid operations, enabling optimal generation scheduling, demand response, and grid stability. Conventional deep learning models such as LSTM, GRU, and CNN-LSTM struggle to simultaneously capture multi-scale temporal dependencies and suppress noise inherent in load time series. This paper proposes WavAttn-BiLSTM, a novel hybrid architecture that integrates Discrete Wavelet Transform (DWT) decomposition, stacked Bidirectional Long Short-Term Memory (BiLSTM) networks, multi-head self-attention, and Temporal Convolutional Networks (TCN) into a unified end-to-end forecasting framework. The DWT decomposes raw load signals into approximation and detail sub-bands, enabling each sub-network to learn frequency-specific temporal patterns. The multi-head attention mechanism adaptively weights the most informative time steps, while the TCN layer captures local convolutional features in the recombined representation. The proposed model is evaluated on three public benchmark datasets: the American Electric Power (AEP) hourly dataset, the ENTSO-E European grid dataset, and the Greek electricity market dataset (2015-2024). WavAttn-BiLSTM achieves MAPE of 0.90%, RMSE of 126.4 MW, and R2 of 0.9805 on the AEP dataset, outperforming eight benchmark models including Transformer, CNN-LSTM, BiLSTM, and ARIMA. Statistical validation using Wilcoxon signed-rank tests (all p < 0.001) and one-way ANOVA (F = 94.6) over 30 independent runs confirms the significance and robustness of these results.
References
[1] U.S. Energy Information Administration (EIA). Electric Power Monthly. Available: https://www.eia.gov/electricity/data.php, 2025.
[2] Ahmad, T., & Zhang, D. (2021). Using the internet of things in smart energy systems and networks. Sustainable Cities and Society, 68(102783), 102783. doi:10.1016/j.scs.2021.102783
[3] Hasanat, S. M., Younis, R., Alahmari, S., Ejaz, M. T., Haris, M., Yousaf, H., … Ullah, Z. (2024). Enhancing load forecasting accuracy in smart grids: A novel parallel multichannel network approach using 1D CNN and Bi‐LSTM models. International Journal of Energy Research, 2024(1). doi:10.1155/2024/2403847
[4] Zhong, B. (2023). Deep learning integration optimization of electric energy load forecasting and market price based on the ANN–LSTM–transformer method. Frontiers in Energy Research, 11. doi:10.3389/fenrg.2023.1292204
[5] Wang, Y., Jiang, W., Wang, C., Song, Q., Zhang, T., Dong, Q., & Li, X. (2023). An electricity load forecasting model based on multilayer dilated LSTM network and attention mechanism. Frontiers in Energy Research, 11(1116465). doi:10.3389/fenrg.2023.1116465
[6] Alsharekh, M. F., Habib, S., Dewi, D. A., Albattah, W., Islam, M., & Albahli, S. (2022). Improving the Efficiency of Multistep Short-Term Electricity Load Forecasting via R-CNN with ML-LSTM. Sensors, 22(18), 6913. https://doi.org/10.3390/s22186913
[7] Alsharekh, M. F., Habib, S., Dewi, D. A., Albattah, W., Islam, M., & Albahli, S. (2022). Improving the efficiency of multistep short-term electricity load forecasting via R-CNN with ML-LSTM. Sensors (Basel, Switzerland), 22(18), 6913. doi:10.3390/s22186913
[8] Buratto, W. G., Muniz, R. N., Nied, A., Barros, C. F. de O., Cardoso, R., & Gonzalez, G. V. (2024). Wavelet CNN‐LSTM time series forecasting of electricity power generation considering biomass thermal systems. IET Generation, Transmission and Distribution, 18(21), 3437–3451. doi:10.1049/gtd2.13292
[9] Xiang, X., Li, X., Zhang, Y., & Hu, J. (2024). A short-term forecasting method for photovoltaic power generation based on the TCN-ECANet-GRU hybrid model. Scientific Reports, 14(1), 6744. doi:10.1038/s41598-024-56751-6
[10] Ait Chaoui, K., EL Fadil, H., Choukai, O., & Ait Omar, O. (2025). A wavelet–attention–convolution hybrid deep learning model for accurate short-term photovoltaic power forecasting. Forecasting, 7(3), 45. doi:10.3390/forecast7030045
[11] Cao, K., Zhang, T., & Huang, J. (2024). Advanced hybrid LSTM-transformer architecture for real-time multi-task prediction in engineering systems. Scientific Reports, 14(1), 4890. doi:10.1038/s41598-024-55483-x
[12] Boucetta, L. N., Amrane, Y., Chouder, A., Arezki, S., & Kichou, S. (2024). Enhanced forecasting accuracy of a grid-connected photovoltaic power plant: A novel approach using hybrid variational mode decomposition and a CNN-LSTM model. Energies, 17(7), 1781. doi:10.3390/en17071781
[13] Sunder, R., Sreeraj, Paul, V., Punia, S. K., Konduri, B., Nabilal, K. V., … Tlija, M. (2024). An advanced hybrid deep learning model for accurate energy load prediction in smart building. Energy Exploration & Exploitation. doi:10.1177/01445987241267822
[14] Hochreiter, S., & Schmidhuber, J. (1997). Long short-term memory. Neural Computation, 9(8), 1735–1780. doi:10.1162/neco.1997.9.8.1735
[15] Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., … Polosukhin, I. (2017). Attention is all you need. doi:10.48550/arXiv.1706.03762
[16] Mallat, S. G. (1989). A theory for multiresolution signal decomposition: the wavelet representation. IEEE Transactions on Pattern Analysis and Machine Intelligence, 11(7), 674–693. doi:10.1109/34.192463
[17] Lea, C., Vidal, R., Reiter, A., & Hager, G. D. (2016). Temporal convolutional networks: A unified approach to action segmentation. In Lecture Notes in Computer Science. Lecture Notes in Computer Science (pp. 47–54). doi:10.1007/978-3-319-49409-8_7
[18] Belletreche, M., Bailek, N., Abotaleb, M., Bouchouicha, K., Zerouali, B., Guermoui, M., … El-Kenawy, E.-S. M. (2024). Hybrid attention-based deep neural networks for short-term wind power forecasting using meteorological data in desert regions. Scientific Reports, 14(1), 21842. doi:10.1038/s41598-024-73076-6
[19] European Network of Transmission System Operators for Electricity (ENTSO-E). Transparency Platform Dataset. Available: https://transparency.entsoe.eu, 2024.
[20] PJM Interconnection LLC. Hourly Load Data - American Electric Power (AEP). Available: https://www.pjm.com/markets-and-operations/energy, 2024.
